Boinvests Holy Grail Playground

[WhoKnows]

Ah, totally missed you have own place here.

See that you have the 'opening on BLUE wave only' implemented.
I Modded Steve's version every new version to do this. Saves a lot of faulty entries... IMHO.
So now I can now use your version.

Unfortunately I am a coding NOOB so adding my 'if (TmaStatus == Tmarangesignal || TmaStatus == TmaInitializing) return(false);' on line 2285ish was about all I could do myself.
Will see what you/we will come up with for further improvements.

I do hope you can also keep track of the improvements Steve implements so that you version will be a HGB+ for power users.

Ok... success met jouw versies,
Peter

BTW if you still have the same core as Steve you may want to look at the TrendWave part for Double entry issues. Yesterday I ran a test on 1H and found it opening a new set of TWbuy and multi's at every 5 minutes in the same H1 candle. I had the wait to next trade at 300.
(1TW + 10 Multi's) * 7 = 77 orders with some even opening and closing in 0.002 seconds.
See picture where I show the entries within the 1H candle by displaying it on a 5 min chart. The 1H chart candle was just one big stack of objects.

Opening TW trades every 5 min within 1Hr time frame.PNG

[WhoKnows]

Regarding timing and changing signals by HGI when HGB reads

I am no coder so can' t do it. But I think we miss a 'signal persistence factor' (SPF) that would be a variable that we can define (so each user can decide on a preference). The SPF would determine how often (and by approximation how long ) a signal must be there before HGB reacts.

I assume HGB only sees if HGI has something in its buffer (no idea how that buffer works... is it an array or a switch/flag) when there is a READ event. The next time when it reads the same signal might be there ...or not.
If it is not it would from a trade perspective be a 'false signal'. The system is correct, but trade-wise you may want to ignore non persistent signals.

What if you can have HGB register if there is a signal in the buffer at the read time and then store that. At the next read time it adds 1 to the number of signals seen before, or resets to zero if there is no signal.
If you then have a selectable SPF persistence factor which determines how many READS the signal must have been in place ....you might have a solution.
Some traders would opt for SPF 1 (as it is now ... only one instance of signal is enough).
Some traders would opt for SPF 5 (the signal must be there for 5 consecutive READS).

Together with the ReadDelay the SPF would by approximation determine how long the signal was present.
E.g. SPF 5 and ReadDelay 60 seconds.... the signal must have been there 5 consecutive times at 1 minute intervals before HGB reacts with a trade.

Hope this makes sense .... I am no coder

[boinvest]

Hope this makes sense .... I am no coder

Yes makes perfect sense. I was thinking about going that road..
Hoping to work on it today.. but didn't get to it yet..

[acostafulano]

Closures seem to be working 100% as intended in 2r.

Also, I'm seeing far less many trades, so I think you got the mindistance thingy to work!

Diego

Fprgot to mention: I'm not having this closures with the SH version. (All opposite closures set to false and all trades closing on either tp or sl as intended).

Maybe has something to do with the modifications?

[acostafulano]

Emiel,

I've been playing with 2R all day, and I can say that the entries are responding very accurately to what the indicator is displaying on the charts.

However, there is one main issue, some of the signals provided by the indi are being missed for some reason,

Please see the screenshot 1:
This is a demo I'm running of instances set to TF M15 (but attached to M5 charts). If you see the trade list, there were not any new trades for the last 20 something minutes or so from when the screenshot was taken.

Screenshot 2: I right away run the "Copy templat to all charts" as a way to "reload" the EA in the charts and right away I get 4 new trades, I inmediately pull the indi to all of these to check if these were in accordance to the signals by the indi on M15 as intended an indeed, they were there. (Screenshots 3-4 as well).

So... in conclusion, I found that the indicator IS picking up quite some trades and those picked up are in accordance to the indis readings, BUT there are some more that are not being picked up instantly, unless one reloads the EA...

I'm attaching my setfile. Everytick ON, BTW.

[boinvest]

So... in conclusion, I found that the indicator IS picking up quite some trades and those picked up are in accordance to the indis readings, BUT there are some more that are not being picked up instantly, unless one reloads the EA...

I'm attaching my setfile. Everytick ON, BTW.

Wow, This is an excellent spot!!
First I was struck by a bit of disbelief.
As I've looked this morning into the calling of the signales.
At this moment, the signals are called on everytick.
This may be the reason people don't see difference in meem and cpu performance.
(yet, have to decide what to do with it)

Dunno, if this is "the thing", but might be in the good direction.
Signals are read everytick, but for trading it takes the HGI readdelay seconds into account.
(Advisably Set to a low value, so you will have multiple checks per candle)

And then there is also a "PostTradeAttemptWaitSeconds" which is default set to 10 minutes.
This might be the one, that not all signals are being evaluated on a 15 min candle.

For the next version I was about to upload, I will change it to a lower value.
Have to see, what I do with this thingy in the future.

[boinvest]

version 2s is attached to post 1.

Couple of minor changes, based on this topic and some PM's.

This include...
Various changes to the order accounting.
Fixed a copy paste error in the calculate takeprofit
Changed the stopLoss routines
Set the PostTradeAttemptWaitSeconds default to 1 Minute.
Some other minor changes which should fix the MinDistanceBetweenTrades.

In the Mod1 version I forgot to place a couple of brackets.
Which caused it to exit on a opposite wavy, even if when you set "closeonopposite TMA" to false.
Finally found that thing!

Have fun!

[acostafulano]

I really think we're getting there.

Once we get it to work the way it's supposed to (which I think you're already pretty darn close!),
I was thinking on using an approach that should be less prone to execution issues, taking advantage on the solid entries but keeping it simple and stupid afterwards:

I'm preparing a post with the approach I'll be using in testing this on my 24/5 home machine, 11-12 demos running with the following common criteria:

- Running M15/H1 or H4 timeframes, but attached to M5 for precision (based on testing this should be irrelevant as attaching 2r to either M5 or M15 for trading on the M15 has yielded me the exact same entries, to the minute)
- Trend,Rad,Wavy, TrendWavy trades only
- all closures on opposite set to FALSE
- Everytick ON
- Max Trades of 6 per type, as to allow multiple entries at any given moment and collect more data
- Min dist between each type of trades: 6

The only differences between this demos will be the following:
SP and TP (or JS)

6670 HGIG - M15 - 40 SL - BEJS 25+2
6674 HGIG - H1 - 40 SL - BEJS 25+2
6682 HGIG - M15 - 60 SL - BEJS 35+2
6686 HGIG - H1 - 60 SL - BEJS 35+2
7204 HGIG - H4 - 55SL-25TP
7208 HGIG - H1 - 45SL-35TP
7212 HGIG - M15 - 45SL-35TP
7216 HGIG - H4 - 45SL-35TP
7220 HGIG - M15 - 55SL-25TP
7224 HGIG - M15 - 40SL-20TP
7228 HGIG - H1 - 55SL-25TP

After 3-6 months, we should be able to analyze the data collected, discard the shitty pairs, the shitty trading hours, and pick the "best" targets for each pair, which I'm sure will vary based on each pair's inherent behavior.

Anyone else is welcome to join me. I can share my setfiles, too.

First thing, is getting the EA working as it should, and then we can take on to the next step!

you rock emiel

So... in conclusion, I found that the indicator IS picking up quite some trades and those picked up are in accordance to the indis readings, BUT there are some more that are not being picked up instantly, unless one reloads the EA...

I'm attaching my setfile. Everytick ON, BTW.

Wow, This is an excellent spot!!
First I was struck by a bit of disbelief.
As I've looked this morning into the calling of the signales.
At this moment, the signals are called on everytick.
This may be the reason people don't see difference in meem and cpu performance.
(yet, have to decide what to do with it)

Dunno, if this is "the thing", but might be in the good direction.
Signals are read everytick, but for trading it takes the HGI readdelay seconds into account.
(Advisably Set to a low value, so you will have multiple checks per candle)

And then there is also a "PostTradeAttemptWaitSeconds" which is default set to 10 minutes.
This might be the one, that not all signals are being evaluated on a 15 min candle.

For the next version I was about to upload, I will change it to a lower value.
Have to see, what I do with this thingy in the future.

[FXe3]

Have you thought about customizing the stoploss for each pair based on ATR? I really like the idea of stacking trades every 20 with profit at 30 but got tired of getting trapped at swing high/lows with 100 pip retraces and no opposite signal to close. I didn't want to use an ATR profit so I calculated ATR and used a 1.5X 4hr ATR as a pip stop.

I really think we're getting there.

Once we get it to work the way it's supposed to (which I think you're already pretty darn close!),
I was thinking on using an approach that should be less prone to execution issues, taking advantage on the solid entries but keeping it simple and stupid afterwards:

I'm preparing a post with the approach I'll be using in testing this on my 24/5 home machine, 11-12 demos running with the following common criteria:

- Running M15/H1 or H4 timeframes, but attached to M5 for precision (based on testing this should be irrelevant as attaching 2r to either M5 or M15 for trading on the M15 has yielded me the exact same entries, to the minute)
- Trend,Rad,Wavy, TrendWavy trades only
- all closures on opposite set to FALSE
- Everytick ON
- Max Trades of 6 per type, as to allow multiple entries at any given moment and collect more data
- Min dist between each type of trades: 6

The only differences between this demos will be the following:
SP and TP (or JS)

6670 HGIG - M15 - 40 SL - BEJS 25+2
6674 HGIG - H1 - 40 SL - BEJS 25+2
6682 HGIG - M15 - 60 SL - BEJS 35+2
6686 HGIG - H1 - 60 SL - BEJS 35+2
7204 HGIG - H4 - 55SL-25TP
7208 HGIG - H1 - 45SL-35TP
7212 HGIG - M15 - 45SL-35TP
7216 HGIG - H4 - 45SL-35TP
7220 HGIG - M15 - 55SL-25TP
7224 HGIG - M15 - 40SL-20TP
7228 HGIG - H1 - 55SL-25TP

After 3-6 months, we should be able to analyze the data collected, discard the shitty pairs, the shitty trading hours, and pick the "best" targets for each pair, which I'm sure will vary based on each pair's inherent behavior.

Anyone else is welcome to join me. I can share my setfiles, too.

First thing, is getting the EA working as it should, and then we can take on to the next step!

you rock emiel

[WhoKnows]

I really think we're getting there.

I'm preparing a post with the approach I'll be using in testing this on my 24/5 home machine, 11-12 demos running with the following common criteria:

- Running M15/H1 or H4 timeframes, but attached to M5 for precision (based on testing this should be irrelevant as attaching 2r to either M5 or M15 for trading on the M15 has yielded me the exact same entries, to the minute)

After 3-6 months,we should be able to analyze the data collected, discard the shitty pairs, the shitty trading hours, and pick the "best" targets for each pair, which I'm sure will vary based on each pair's in

One thing I have been wrecking my brain over is the basic question if it is wise to have the same set for every pair.

In one of the other threads Feucht shared some long term results. http://www.stevehopwoodforex.com/phpBB3 ... att_sr=att
- Some pairs proved to be worse than others. Do these pairs need a different setting? Or why are they consistently worse? Do we need different settings for these pairs to account for different volatility?
- Or are some pairs better on certain time frames?

The main conclusion that I drew from Feucht's results was that there is a definitive time effect of when the trades were started.

Complications and decisions
A scientific background and curiosity how things work is probably a negative in trading.
I should be focusing on just running with using all pairs, be in the market, and make money


It will be interesting to follow your results. How do you manage so many instances?
Must be a lot of work.